- Email: [email protected]
Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam
Accepted Manuscript Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam Chi-Hsien Huang, Ching-Yuan Su, Takeru Okada, Lain-Jong Li, Kuan-I Ho, Pei-Wen Li, Inn-Hao Chen, Chien Chou, Chao-Sung Lai, Seiji Samukawa PII: DOI: Reference:
S0008-6223(13)00408-9 http://dx.doi.org/10.1016/j.carbon.2013.04.099 CARBON 8021
To appear in:
Carbon
Received Date: Accepted Date:
10 January 2013 30 April 2013
Please cite this article as: Huang, C-H., Su, C-Y., Okada, T., Li, L-J., Ho, K-I., Li, P-W., Chen, I-H., Chou, C., Lai, C-S., Samukawa, S., Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam, Carbon (2013), doi: http://dx.doi.org/10.1016/j.carbon.2013.04.099
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam
Chi-Hsien Huanga, Ching-Yuan Sua, Takeru Okadab, Lain-Jong Lic, Kuan-I Hoa, Pei-Wen Lid, Inn-Hao Chend, Chien Choue, Chao-Sung Laia,*, Seiji Samukawab,*
a
Department of Electronic Engineering, Chang Gung University, 259 Wen-Hwa 1st
Road, Kwei-Shan Tao-Yuan, 333 (Taiwan) b
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai
980-8577, Japan c
Institute of Atomic and Molecular Sciences, Academia Sinica, Taiwan, Taipei 10617,
Taiwan d
Department of Electrical Engineering, National Central University, Tao-Yuan 320,
Taiwan e
Graduate Institute of Electro-Optical Engineering, Chang Gung University, Tao-Yuan
333, Taiwan
*Corresponding Authors: Tel: +886-3-2118800 Fax: +886-3-2118507, [email protected] (Chao-Sung Lai); Tel/Fax: +81-22-2175240, Email: [email protected] (Seiji Samukawa)
Abstract Top-down process, comprising lithography and plasma etching is widely used in very-large-scale integration due to its scalability, has the greatest potential to fabricate graphene nanoribbon based nanoelectronic devices for large-scale intergraded circuits. However, conventional plasma etching inevitably introduces plenty of damage or defects to the etched materials, which drastically degrades the performance of nano materials. In this study, extremely low-damage neutral beam etching (NBE) is applied to
fabricate
ultra-low-defect
graphene
nanoribbon
array
(GNR).
The
ultra-low-edge-defect GNRs are fabricated by E-beam lithography followed by oxygen NBE from large-scale chemical-vapor-deposition-grown graphene. AFM images clearly shows the GNRs patterned by NBE and E-beam lithography, and Raman spectroscopy exhibits extremely low ID/IG of GNRs, which indicate that high-quality GNRs can be successfully fabricated by neutral beam. We also demonstrated bottom-gated field-effect transistor with the high-quality GNR and observed a high carrier mobility (>200 cm2 V−1 s−1) at room temperature.
1. Introduction Since it was first reported in 2004 [1], graphene, a two-dimensional, single layer of sp2 hybridized carbon atoms, has received tremendous attention and research interest for its fundamental physics and for potential applications because it possesses unique characteristics such as a linear energy dispersion relation and exceptionally high electronic conductivity. Graphene is certainly a very promising material for future nanoelectronic devices [2-4]. However, the fact that it is a semimetal with zero-bandgap and retains a high conductivity even at the charge neutrality point means its potential application in field-effect transistors (FETs) operating at room temperature is limited [5-7]. Graphene nanoribbon (GNR), a quasi-one-dimensional graphene nanostructure, has an effective bandgap (Eg) because of the lateral confinement of charge carriers and circumvent some problems encountered by the gapless band structure of 2D graphene [8]. Studies have also confirmed that the Eg of the GNRs inversely scales inversely with the nanoribbon width [8,9]. Several approaches have been used to fabricate or synthesize GNRs, including chemical procedures [9-11], unzipping of carbon nanotubes [12-14], and top-down lithographic patterning [15-17]. Among existing approaches, top-down lithographic patterning, i.e., lithographic patterning followed by top-down oxygen (O2) plasma etching, is very attractive for fabricating well-arranged GNRs required for large-scale device
integration. However, conventional O2 plasma etching always produces a large number of defects on the edges of GNRs [16-19]. It was reported that the disordered edge is around 5 nm in width [16,18], which indicates that the defects in graphene dominate the electrical characteristics of GNR-based FETs. Therefore, it has proven to be difficult to obtain a large Eg and sufficient mobility for room-temperature operation of GNR-based FETs using wide GNRs (>10 nm) [15,20]. In addition, most GNRs to date have been fabricated from exfoliated graphene [8,21,22] and not from wafer-scale graphene, such as chemical vapor deposition (CVD)-grown graphene. Here, we report GNRs lithographically patterned by electron beam lithography (EBL) followed by O2 neutral beam etching (NBE) of large-scale CVD-grown graphene. In contrast to conventional O2 plasma etching, NBE is a top-down etching technique imposing ultra-low defects and produces high-quality GNRs. The fabricated GNR arrays were visualized using optical microscopy (OM) and atomic force microscopy (AFM), and evidence of the high quality of the GNRs with ultra-low defects at the edges was confirmed by Raman spectroscopy. We also demonstrated a bottom-gate transistor with a GNR array, which showed a high carrier mobility at room temperature.
2. Experimental 2.1 GNR Preparation The GNR sample preparation started with the growth of a large-sized monolayer graphene film on copper foil using CVD in a tubular quartz furnace [23]. A Copper foil was placed at the center of the quartz tube, and the system was flushed with a constant flow of hydrogen (50 sccm) at 760 mTorr for 50 min. The Cu foil was annealed at 1000 °C for 40 min to remove organic matter and oxides from the surface. A gas mixture of methane and hydrogen (CH4 = 60 sccm and H2 = 15 sccm at 750 mTorr) was introduced into the system at 1000 °C for the growth of the graphene layer. After the graphene films were grown, the graphene and Cu foil were cooled to 25 °C. The as-grown graphene was transferred from the Cu foil to a thermally grown SiO2 (300 nm thick) substrate atop a heavily p-doped Si substrate (SiO2/Si) using the following transfer procedure: (i) a poly-(methyl methacrylate) (PMMA) film was spin-coated onto the surface of the graphene on the Cu foil; (ii) the PMMA/graphene layer was separated from the Cu foil by chemical etching of the Cu in an iron chloride solution; (iii) the suspended PMMA–graphene layer was placed on the surface of deionized water overnight to remove any residual Cu etchant; (iv) the PMMA/graphene layer was then transferred to the SiO2/Si substrate; (v) after drying on a hotplate at 120 °C, the PMMA was dissolved and washed away by soaking the
substrate in acetone to leave only the monolayer graphene film on the SiO2/Si substrate; (vi) finally, the sample was rinsed with isopropyl alcohol and deionized water to obtain graphene/SiO2/Si. The GNRs were prepared by EBL patterning followed by O2 NBE with the following conditions: chamber pressure of 0.7 mTorr, an RF power of 250 W, an O2 flow rate of 6.7 scm, and an irradiation time of 3.5 min. A negative-tone resist, hydrogen silsesquioxae (HSQ), with a thickness of around 150 nm was spin-coated onto the graphene/SiO2/Si samples. Following the patterning and subsequent development of the etching mask pattern, the unmasked graphene area was etched by O2 NBE. After removing the HSQ by using diluted hydrofluoric acid, we obtained GNRs on the SiO2/Si substrate. Four GNR arrays with widths of 30, 50, 70, and 100 nm were fabricated, and their length was around 8 μm. A total of 10–15 nanoribbons of the same width were fabricated in an array spanning 2–3 μm to ensure a sufficient Raman signal for the analysis. One large graphene sheet with a square shape (width: 5 μm ) was also fabricated as reference bulk graphene. 2.2 Characterization of GNR Tapping mode of Atomic Force Microscope (AFM) was used to measure the profile of the fabricated graphene nanoribbon arrays. Raman spectra were collected in a NT-MDT confocal Raman microscopic system (laser wavelength: 473 nm; laser spot size: ~0.5 μm; accumulation time: 10 s). The Si peak at 520 cm−1 was used as a
reference for the wavenumber calibration before each measurement. 2.3 Electrical Measurement To evaluate the electrical characteristics of GNRs, bottom-gate FETs having GNR arrays (width: 70 nm) with 10 nanoribbons, prepared by EBL patterning followed by O2 NBE, were fabricated. The 70-nm GNR array was covered with a Cu grid (as a hard mask), and Cr/Au (2 nm/60 nm) source and drain electrodes were subsequently deposited on the GNR array by a thermal evaporator to form the FET. The FET channel was 5 μm in length, and the width of the source and drain electrodes was 12 μm. All electrical measurements were conducted under ambient conditions using a Keithley-4200 semiconductor analyzer.
3. Results and discussion 3.1 Neutral Beam Top-down process comprising lithography and plasma etch is widely used in semiconductor manufacturing for patterning. However, plasma etch is indispensible to damage the etched materials resulting from high energy ion bombardment and ultraviolet (UV) irradiation generated in plasma [24,25]. Especially, UV photons can penetrate into the materials to induce defects with depth of tens of nanometers because of the energy of UV photons higher than bond energy of material. Therefore,
an ultra low damage etch method is highly demanded to pattern materials into nanometer scale. Neutral beam developed by Prof. Samukawa’s group is a promising candidate for such severe request [26-28]. The neutral beam etching (NBE) system shown in Figure 1 consists of plasma and process chambers that are separated by a graphite plate with many apertures. The sample is placed on the stage in the process chamber. Inductive coupled plasma (ICP) is generated by applying RF power to 3-turn antenna in the plasma chamber. The charged particles can be effectively neutralized by collision with the sidewall of the apertures when passing through them to the process chamber. The neutralized efficiency can be as high as more than 95%. On the other hand, more than 90% of UV intensity can be eliminated when UV photons pass through the graphite plate to the process chamber. As a result, only energetic oxygen neutral beams bombard the graphene surface for nanoribbon patterning. The beam energy ranging from several to hundreds of electron-volts can be controlled by applying an RF or a DC bias to the bottom electrode. Additionally, as the neutral beams are well aligned, the mask pattern can be transferred precisely to the graphene sheet. To demonstrate the potential of low damage of neutral beam to graphene material, pristine graphene without pattern were irradiated by nitrogen (N2) NB and N2 plasma with the same ICP conditions as follows: chamber pressure of 1.7 mTorr, an RF power
of 250 W, an Ar flow rate of 6.7 sccm, and an irradiation time of 5 min. The generated ion and beam flux under these conditions are approximately 1016 s-1 cm-2 for N2 NB and N2 plasma, respectively. Figure 2 shows the Raman spectra of graphene sheet after irradiations of N2 NB and N2 plasma. Well-defined features of the Raman spectra can be identified: G-band (in-plane vibration of sp2 carbon atoms, ~1585 cm−1); the D-band (defect induced bands, ~1350 cm−1); and the 2D-band (double-resonance band, ~2700 cm−1). We can clearly see that no featured bands of graphene can be observed after irradiation of N2 plasma. On the contrary, the Raman spectrum of the graphene sheet after irradiation of N2 NB shows the featured bands of graphene (G-band and 2D-band) with small D-band. Those results indicate that the NB process is an ultra low damage process for graphene due to low energy ion bombardment and elimination of UV irradiation. 3.2 Structure of Graphene Nanoribbon Array Four nanoribbon arrays with widths of 30, 50, 70, and 100 nm were patterned by EBL and subsequently etched by O2 NBE to obtain GNR arrays (see Experimental Section for details). Figure 3(a) shows the OM images of the GNR arrays after removing hydrogen silsesquioxane (HSQ), a negative-tone resist. We found that the unmasked graphene was etched away cleanly by NBE and only the GNR arrays were left, which clearly have a rectangular shape. Figure 3(b) presents the AFM images of GNR arrays
of various widths after removing HSQ. The AFM images revealed that the GNR arrays can be fabricated uniformly on a large scale using EBL followed by O2 NBE. The profiles of the GNR arrays with various nanoribbon widths measured by AFM are shown in Figure 3(c). The SiO2 under the graphene is also etched when removing the HSQ by using diluted hydrofluoric acid, the height is normalized by the total depth (around 30 nm) as measured by AFM. We can clearly see that the widths of the fabricated GNR arrays are close to the same size as the patterns designed by EBL. The images and figures exhibit uniform patterning of the GNR arrays using EBL followed by O2 NBE for large-scale integrated circuits. 3.3 Raman Spectroscopy of GNR Representative Raman spectra of the GNR arrays are shown in Figure 4(a). Well-defined features of the Raman spectra can be identified as mentioned in section 2.1 and an additional D′-band was observed. Both D- and D′-band are responsible for defect-induced bands and D′-band appears when the GNR width is narrower due to the presence of edges. [16,19,29]. A square graphene sheet with a width of 5 μm shows a symmetric peak with a high intensity (the intensity ratio of the 2D-band to the G-band is higher than ~4) with an almost invisible D-band. These features indicate that high-quality and single-layer graphene was grown by CVD [30]. As shown in Figure 4(a), clear D- and D′-bands were observed because of defect generation at the
GNR edge; no such bands were observed in the bulk sample. No substantial shift in the spectra was observed as the width of the GNR array changed. Figure 4(b) shows the ratio of the integrated intensities of the D- and G-bands, denoted as ID and IG respectively, as a function of the GNR width. The ID/IG was the indicator of the defect level at the GNR edge. Although the ID/IG ratio increases as the GNR width decreases, the ratio is impressively lower than that of the GNRs fabricated by conventional O2 plasma etching [7,16]. Since the graphene is covered by resist, the anisotropic ion bombardment cannot induce such wide edge defect in GNR under the resist. Therefore, the low edge defect should arise from the reduction of UV irradiation damage. Because the UV photons, which are isotropic, can penetrate into the resist and reach graphene nanoribbon especially from the edge where penetration path is relatively short as schematically shown in Figure 5. It was reported that the photon energy of O2 plasma is as high as 9.5 eV [31] which is much higher than graphene C-C bond energy of 4.9 eV [32]. Therefore, in the case of conventional plasma etch, the sp2 C-C bonds at the edge of GNR are destroyed to form wide range of defects (indicated as w in Figure 5(a)) observed as high ID/IG in Raman spectroscopy. On the contrary, the UV photons are almost eliminated in the NB system and therefore only narrow range of defect (as indicated as w′ in Figure 5(b)), contributing to small ID/IG, is induced owing to the C-C bond breakage at the edge resulting from NB
bombardment, which is the nature of etch process. Moreover, the ratio is comparable to that of GNRs prepared by unzipping carbon nanotubes followed by post annealing [13,14], a procedure which is not suitable for large-scale fabrication. It has been reported that post-fabrication treatment is required to repair the high degree of disorder at the edge [14,17,33], and this is what makes the unzipping procedure complicated. The results in Figure 3 strongly suggest that high-quality GNR arrays can be fabricated easily using NBE without post-fabrication treatment. Furthermore, Raman mapping of the etched samples, as shown in Figure 6, further demonstrated the potential of large-scale GNRs fabrication using NBE; no graphene-feature bands were observed in the Raman spectra of the unmasked areas. 3.4 Characteristics of GNR-based FET To evaluate the electrical characteristics of the GNRs fabricated by NBE, a bottom-gate FET with a GNR array of 10 nanoribbons (width: 70 nm) was fabricated (shown schematically in Figure 7(a); an OM image of the FET is shown in Figure 7(b). The transfer curve (drain current Id vs. gate voltage Vg) with an applied voltage between the source and the drain (Vsd) of 0.5 V for the FET is shown in Figure 6c. We repeated the FET fabrication for three times and measured 15 devices. We obtained 10 devices with similar transfer curve as shown in Figure 7(c). The field-effect mobility (μ) was extracted based on the slope ΔId/ΔVg of a line fitted to the linear regime of the
transfer curve using the following formula [21,34,35]: μ=
(ΔI
d
ΔV g )L
nWV sd C g
(1)
where L, n, and Cg are the channel length, number of GNRs, and capacitance per unit area between the GNR and the bottom gate, respectively. We obtained an effective field-effect mobility of more than 200 cm2 V−1 s−1. Although the mobility value is not as high as single crystalline graphene [21,22] that is obtained by mechanical exfoliated method. However, mechanical exfoliation method to obtain graphene is not suitable for very-large-scale integration (VLSI), which is highly anticipated application of graphene, because it is done by hand and the area of graphene is only around several hundreds of micrometer square. This small area of single crystalline graphene sheet is without grain boundaries and different from large are of polycrystalline CVD-grown graphene. Therefore, the mobility of more than 200 cm2 V−1 s−1 is high for CVD-grown GNR due to that CVD-grown graphene is inevitable to contain grain boundaries in channel region, especially in long channel (5μm in this study). Comparing to mechanical exfoliation method, CVD is a VLSI-compatible process, and the area of CVD-grown graphene sheet can reach wafer-size. The result demonstrates high-quality GNRs with low edge defects, which reduces edge scattering for high carrier mobility. Furthermore, we observed high Ion/Ioff current ratio ranging from 103 to 104 at room temperature in our experiment, which is much higher
than other reported results [3]. Such large Ion/Ioff current ratio should contribute from transport gap. M. Y. Han et al. reported that transport gap strongly depends on length of GNR [36]. Since our fabricated GNR is 8μm long (i.e. long channel), which is much longer than that fabricated by other reported approaches. As a result, we considered that such large Ion/Ioff current ratio may result from long channel effect.
4. Conclusions In summary, promising GNR arrays with low edge-defects were demonstrated using EBL and followed by O2 NBE on large-scale CVD-grown graphene. AFM and Raman analysis show that high-quality GNR arrays can be fabricated because NB system can provide low energy and well-controlled neutral beams without UV irradiation damage. A bottom-gated FET having GNR array with nanoribbon width of 70 nm was fabricated. We obtained a high field-effect mobility of ~200 cm2 V−1 s−1 and observed a high Ion/Ioff ratio (103~104) in our fabricated FET. This method is a good candidate for fabricating high-performance GNR-based electronic devices for large-scale integrated circuits.
References
[1] Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV, Firsov AA. Electric field effect in atomically thin carbon film. Science 2004;306(5696):666-9. [2] Geim AK. Graphene: status and prospects. Science 2009;324(5934):1530-4. [3] Schwierz F. Graphene transistors. Nat Nanotech 2010;5(7):487-96. [4] Lim WS, Kim YY, Kim H, Jang S, Kwon N, Park BJ, Ahn JH, Chung I, Hong BH, Yeom GY. Atomic layer etching of graphene for full graphene fabrication. Carbon 2012;50(2):429-35. [5] Son YW, Cohen ML, Louie SG. Energy gaps in graphene nanoribbons. Phys Rev Lett. 2006;97(21):216803. [6] Lee H, Kim S, Ihm J, Son YW, Choi HJ. Field-induced recovery of massless Dirac fermions in epitaxial graphene on SiC. Carbon 2011;49(7):2300-5. [7] Wang X, Dai H. Etching and narrowing of graphene from the edges. Nat Chem 2010;2(8):661-5. [8] Han MY, Özyilmaz B, Zhang Y, Kim P. Energy band-gap engineering of graphene nanoribbons. Phys Rev Lett 2007;98(20):206805. [9] Li X, Wang X, Zhang L, Lee S, Dai H. Chemically derived ultrasmooth graphene nanoribbon semiconductors. Science 2008;319(5867):1229-32.
[10] Yang X, Dou X, Rouhanipour A, Zhi L, Räder HJ, Müllen K. Two-dimensional graphene nanoribbons. J Am Chem Soc 2008;130(13):4216-7. [11] Cao X, He Q, Shi W, Li B, Zeng Z, Shi Y, Yan Q, Zhang H. Graphene oxide as a carbon
source
for
controlled
growth
of
carbon
nanowires.
Small
2011;7(9):1199-202. [12] Kosynkin DV, Higginbotham AL, Sinitskii A, Lomeda JR, Dimiev A, Price BK, Tour JM. Longitudinal unzipping of carbon nanotubes to form graphene nanoribbons. Nature 2009;458(7240):872-6. [13] Cataldo F, Compagnini G, Patané G, Ursini O, Angelini G, Ribic PR, Margaritondo G, Cricenti A, Palleschi G, Valentini F. Graphene nanoribbons produced by the oxidative unzipping of single-wall carbon nanotubes. Carbon 2010;48(9):2596-602. [14] Shimzu T, Haruyama J, Marcano DC, Kosinkin DV, Tour JM, Hirose K, Suenaga K. Large intrinsic energy bandgaps in annealed nanotube-derived graphene nanoribbons. Nat Nanotech 2011;6(1):45-50. [15] Bai J, Duan X, Huang Y. Rational fabrication of graphene nanoribbons using nanowire etching mask. Nano Lett 2009;9(5):2083-7. [16] Ryu S, Maultzsch J, Han MY, Kim P, Brus LE. Raman spectroscopy of lithographically patterned graphene nanoribbons. ACS Nano 2011;5(5):4123-30.
[17] Liu L, Zhang Y, Wang W, Gu C, Bai X, Wang E. Nanosphere lithography for the fabriation of ultranarrow graphene nanoribbons and on-chip bandgap tunning of graphene. Adv Mater 2011;23(10):1246-51. [18] Lin MW, Ling C, Zhang Y, Yoon HJ, Cheng MC, Agapito LA, Kioussis N, Widjaja N, Zhou Z. Room-temperature high on/off ratio in suspended graphene nanoribbon field-effect transistors. Nanotechnology 2011;22(26):265201. [19] Bischoff D, Güttinger J, Dröscher S, Ihn T, Ensslin K, Stampfer C. Raman spectroscopy on etched graphene nanoribbons. J Appl Phys 2011;109(7):073710. [20] Yang Y, Murali R. Impact of size effect on graphene nanoribbon transport. IEEE Electron Dev Lett 2010;31(3):237. [21] Shin YS, Son JY, Jo MH, Shin YH, Jang HM. High-mobility graphene nanoribbons prepared using polystyrene dip-pen nanolithography. J Am Chem Soc 2011;133(15):5623-5. [22] Liao L, Bai J, Lin YC, Qu Y, Huang Y, Duan X. High-performance top-gated graphene-nanoribbon
transistors
using
zirconium
oxide
nanowire
as
high-dielectric-constant gate dielectrics. Adv Mater 2010;22(17):1941-5. [23] Lu AY, Wei SY, Wu CY, Hernandez Y, Chen TY, Liu TH, Pao CW, Chen FR, Li LJ, Juang ZY. Decoupling of CVD graphene by controlled oxidation of recrystallized Cu. RSC Adv 2012;2(7):3008-13.
[24] Ohchi T, Kobayashi S, Fukasawa M, Kugimiya K, Kinoshita T, Takizawa T, Hamaguchi S, Kamide Y, Tatsumi T. Reducing damage to Si substrate during gate etching process. Jpn J Appl Phys 2008;47(7):5324-6. [25] Ishikawa Y, Okigawa M, Yamazaki S, Samukawa S. Reduction of palsma-indued damage in SiO2 film during pulse-time-modulated plasma irradiation. J Vac Sci Technol B 2005;23(2):389-94. [26] Samukawa S, Sakamoto K, Ichiki K. High-efficiency neutral-beam generation by combination of inductively coupled plasma and parallel plate DC bias. Jpn J Appl Phys 2001;40(7B):L779. [27] Samukawa S, Sakamoto K, Ichiki K. Generation of high-efficiency neutral beam by using negative ions in an inductively coupled plasma source. J Vac Sci Tech A 2002;20(5):1566-73. [28] Huang CH, Wang XY, Igarashi M, Murayama A, Okada Y, Yamashita I, Samukawa S. Optical absorption characteristics of highly ordered and dense two-dimensional array of silicon nanodiscs. Nanotechnology 2011;22():105301. [29] Eckmann A, Felten A, Mishchenko A, Britnell L, Krupke R, Novoselov KS, Casirashi C. Probing the nature of defects in graphene by Raman spectroscopy. Nano Lett 2012;12(8):3925-30.
[30] Graf D, Molitor F, Ensslin K, Stampfer C, Jungen A, Hierold C, Wirtz L. Spatially resolved Raman spectroscopy of single- and few-layer graphene. Nano Lett 2007;7(2):238-42. [31] Ishikawa Y, Ichihashi Y, Yamasaki S, Samukawa S. UV photon-induced defects and its control in plasma etching process. J Appl Phys 2008;104(6):063308. [32] Erickson K, Erni R, Lee Z, Alem N, Gannett W, Zettl A. Determination of the local chemical structure of graphene oxide and reduced graphene oxide. Adv Mater 2010;22(40):4467-72. [33] Jiao L, Zhang L, Ding L, Liu J, Dai H. Aligned graphene and crossbars from unzipped carbon nanotubes. Nano Res 2010;3(6):387-94. [34] Lee CW, Weng CH, Wei L, Chen Y, Chan-Park MB, C. H. Tsai, K. C. Leou, C. H. P. Poa, J. Wang and, L. J. Li. Toward high-performance solution-processed carbon nanotube network transistors by removing nanotube bundles. J. Phys. Chem. C, 2008, 112(32), 12089-91. [35] Lin MW, Ling C, Agapito LA, Kioussis N, Zhang Y, Cheng MMC, Wang WL, Kaxiras E, Zhou Z. Approaching the intrinsic band gap in suspended high-mobility graphene nanoribbons. Phys Rev B 2011;84(12):125411. [36] Han MY, Brant JC, Kim P. Electron transport in disordered graphene nanoribbons. Phys Rev Lett 2010;104(5):056801.
Figures
Figure 1. Neutral beam system. ICP plasma is generated by applying RF power to 3-turn antenna. Charged species are neutralized and UV photons are eliminated by graphite plate with many apertures. Only neutral beams with controllable energy arrive at the etch surface in etch chamber.
Figure 2 Raman spectra of graphene sheet irradiated by N2 plasma and N2 neutral beam under the same ICP conditions.
Figure 3 a) Optical microscope images, b) atomic force microscope images and c) profiles of the GNR arrays with various nanoribbons.
Figure 4. a) Representative Raman spectra of GNR arrays with various widths. b) Ratio of the integrated intensities of the D- and G-bands (ID/IG) as a function of the
GNR width. Results are shown for GNRs prepared by our method and other approaches.
Figure 5. Illustration of defect generation near the edge of GNR patterned by a) conventional plasma and b) neutral beam.
Figure 6. Raman mapping of the etched graphene sheet. Position A is inside the GNR array (GNR width: 70nm), and positions B and C are outside the array. The Raman mapping demonstrates uniform patterning using EBL followed by O2 NBE.
Figure 7 a) Structure of the field-effect transistor with a GNR array. b) Optical microscope image of the bottom-gate FET device. c) The drain current vs. gate voltage (Id vs. Vg) of the FET with a GNR array of 10 nanoribbons (width: 70 nm). (Vsd = 0.5 V)
S0008-6223(13)00408-9 http://dx.doi.org/10.1016/j.carbon.2013.04.099 CARBON 8021
To appear in:
Carbon
Received Date: Accepted Date:
10 January 2013 30 April 2013
Please cite this article as: Huang, C-H., Su, C-Y., Okada, T., Li, L-J., Ho, K-I., Li, P-W., Chen, I-H., Chou, C., Lai, C-S., Samukawa, S., Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam, Carbon (2013), doi: http://dx.doi.org/10.1016/j.carbon.2013.04.099
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Ultra-low-edge-defect graphene nanoribbons patterned by neutral beam
Chi-Hsien Huanga, Ching-Yuan Sua, Takeru Okadab, Lain-Jong Lic, Kuan-I Hoa, Pei-Wen Lid, Inn-Hao Chend, Chien Choue, Chao-Sung Laia,*, Seiji Samukawab,*
a
Department of Electronic Engineering, Chang Gung University, 259 Wen-Hwa 1st
Road, Kwei-Shan Tao-Yuan, 333 (Taiwan) b
Institute of Fluid Science, Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai
980-8577, Japan c
Institute of Atomic and Molecular Sciences, Academia Sinica, Taiwan, Taipei 10617,
Taiwan d
Department of Electrical Engineering, National Central University, Tao-Yuan 320,
Taiwan e
Graduate Institute of Electro-Optical Engineering, Chang Gung University, Tao-Yuan
333, Taiwan
*Corresponding Authors: Tel: +886-3-2118800 Fax: +886-3-2118507, [email protected] (Chao-Sung Lai); Tel/Fax: +81-22-2175240, Email: [email protected] (Seiji Samukawa)
Abstract Top-down process, comprising lithography and plasma etching is widely used in very-large-scale integration due to its scalability, has the greatest potential to fabricate graphene nanoribbon based nanoelectronic devices for large-scale intergraded circuits. However, conventional plasma etching inevitably introduces plenty of damage or defects to the etched materials, which drastically degrades the performance of nano materials. In this study, extremely low-damage neutral beam etching (NBE) is applied to
fabricate
ultra-low-defect
graphene
nanoribbon
array
(GNR).
The
ultra-low-edge-defect GNRs are fabricated by E-beam lithography followed by oxygen NBE from large-scale chemical-vapor-deposition-grown graphene. AFM images clearly shows the GNRs patterned by NBE and E-beam lithography, and Raman spectroscopy exhibits extremely low ID/IG of GNRs, which indicate that high-quality GNRs can be successfully fabricated by neutral beam. We also demonstrated bottom-gated field-effect transistor with the high-quality GNR and observed a high carrier mobility (>200 cm2 V−1 s−1) at room temperature.
1. Introduction Since it was first reported in 2004 [1], graphene, a two-dimensional, single layer of sp2 hybridized carbon atoms, has received tremendous attention and research interest for its fundamental physics and for potential applications because it possesses unique characteristics such as a linear energy dispersion relation and exceptionally high electronic conductivity. Graphene is certainly a very promising material for future nanoelectronic devices [2-4]. However, the fact that it is a semimetal with zero-bandgap and retains a high conductivity even at the charge neutrality point means its potential application in field-effect transistors (FETs) operating at room temperature is limited [5-7]. Graphene nanoribbon (GNR), a quasi-one-dimensional graphene nanostructure, has an effective bandgap (Eg) because of the lateral confinement of charge carriers and circumvent some problems encountered by the gapless band structure of 2D graphene [8]. Studies have also confirmed that the Eg of the GNRs inversely scales inversely with the nanoribbon width [8,9]. Several approaches have been used to fabricate or synthesize GNRs, including chemical procedures [9-11], unzipping of carbon nanotubes [12-14], and top-down lithographic patterning [15-17]. Among existing approaches, top-down lithographic patterning, i.e., lithographic patterning followed by top-down oxygen (O2) plasma etching, is very attractive for fabricating well-arranged GNRs required for large-scale device
integration. However, conventional O2 plasma etching always produces a large number of defects on the edges of GNRs [16-19]. It was reported that the disordered edge is around 5 nm in width [16,18], which indicates that the defects in graphene dominate the electrical characteristics of GNR-based FETs. Therefore, it has proven to be difficult to obtain a large Eg and sufficient mobility for room-temperature operation of GNR-based FETs using wide GNRs (>10 nm) [15,20]. In addition, most GNRs to date have been fabricated from exfoliated graphene [8,21,22] and not from wafer-scale graphene, such as chemical vapor deposition (CVD)-grown graphene. Here, we report GNRs lithographically patterned by electron beam lithography (EBL) followed by O2 neutral beam etching (NBE) of large-scale CVD-grown graphene. In contrast to conventional O2 plasma etching, NBE is a top-down etching technique imposing ultra-low defects and produces high-quality GNRs. The fabricated GNR arrays were visualized using optical microscopy (OM) and atomic force microscopy (AFM), and evidence of the high quality of the GNRs with ultra-low defects at the edges was confirmed by Raman spectroscopy. We also demonstrated a bottom-gate transistor with a GNR array, which showed a high carrier mobility at room temperature.
2. Experimental 2.1 GNR Preparation The GNR sample preparation started with the growth of a large-sized monolayer graphene film on copper foil using CVD in a tubular quartz furnace [23]. A Copper foil was placed at the center of the quartz tube, and the system was flushed with a constant flow of hydrogen (50 sccm) at 760 mTorr for 50 min. The Cu foil was annealed at 1000 °C for 40 min to remove organic matter and oxides from the surface. A gas mixture of methane and hydrogen (CH4 = 60 sccm and H2 = 15 sccm at 750 mTorr) was introduced into the system at 1000 °C for the growth of the graphene layer. After the graphene films were grown, the graphene and Cu foil were cooled to 25 °C. The as-grown graphene was transferred from the Cu foil to a thermally grown SiO2 (300 nm thick) substrate atop a heavily p-doped Si substrate (SiO2/Si) using the following transfer procedure: (i) a poly-(methyl methacrylate) (PMMA) film was spin-coated onto the surface of the graphene on the Cu foil; (ii) the PMMA/graphene layer was separated from the Cu foil by chemical etching of the Cu in an iron chloride solution; (iii) the suspended PMMA–graphene layer was placed on the surface of deionized water overnight to remove any residual Cu etchant; (iv) the PMMA/graphene layer was then transferred to the SiO2/Si substrate; (v) after drying on a hotplate at 120 °C, the PMMA was dissolved and washed away by soaking the
substrate in acetone to leave only the monolayer graphene film on the SiO2/Si substrate; (vi) finally, the sample was rinsed with isopropyl alcohol and deionized water to obtain graphene/SiO2/Si. The GNRs were prepared by EBL patterning followed by O2 NBE with the following conditions: chamber pressure of 0.7 mTorr, an RF power of 250 W, an O2 flow rate of 6.7 scm, and an irradiation time of 3.5 min. A negative-tone resist, hydrogen silsesquioxae (HSQ), with a thickness of around 150 nm was spin-coated onto the graphene/SiO2/Si samples. Following the patterning and subsequent development of the etching mask pattern, the unmasked graphene area was etched by O2 NBE. After removing the HSQ by using diluted hydrofluoric acid, we obtained GNRs on the SiO2/Si substrate. Four GNR arrays with widths of 30, 50, 70, and 100 nm were fabricated, and their length was around 8 μm. A total of 10–15 nanoribbons of the same width were fabricated in an array spanning 2–3 μm to ensure a sufficient Raman signal for the analysis. One large graphene sheet with a square shape (width: 5 μm ) was also fabricated as reference bulk graphene. 2.2 Characterization of GNR Tapping mode of Atomic Force Microscope (AFM) was used to measure the profile of the fabricated graphene nanoribbon arrays. Raman spectra were collected in a NT-MDT confocal Raman microscopic system (laser wavelength: 473 nm; laser spot size: ~0.5 μm; accumulation time: 10 s). The Si peak at 520 cm−1 was used as a
reference for the wavenumber calibration before each measurement. 2.3 Electrical Measurement To evaluate the electrical characteristics of GNRs, bottom-gate FETs having GNR arrays (width: 70 nm) with 10 nanoribbons, prepared by EBL patterning followed by O2 NBE, were fabricated. The 70-nm GNR array was covered with a Cu grid (as a hard mask), and Cr/Au (2 nm/60 nm) source and drain electrodes were subsequently deposited on the GNR array by a thermal evaporator to form the FET. The FET channel was 5 μm in length, and the width of the source and drain electrodes was 12 μm. All electrical measurements were conducted under ambient conditions using a Keithley-4200 semiconductor analyzer.
3. Results and discussion 3.1 Neutral Beam Top-down process comprising lithography and plasma etch is widely used in semiconductor manufacturing for patterning. However, plasma etch is indispensible to damage the etched materials resulting from high energy ion bombardment and ultraviolet (UV) irradiation generated in plasma [24,25]. Especially, UV photons can penetrate into the materials to induce defects with depth of tens of nanometers because of the energy of UV photons higher than bond energy of material. Therefore,
an ultra low damage etch method is highly demanded to pattern materials into nanometer scale. Neutral beam developed by Prof. Samukawa’s group is a promising candidate for such severe request [26-28]. The neutral beam etching (NBE) system shown in Figure 1 consists of plasma and process chambers that are separated by a graphite plate with many apertures. The sample is placed on the stage in the process chamber. Inductive coupled plasma (ICP) is generated by applying RF power to 3-turn antenna in the plasma chamber. The charged particles can be effectively neutralized by collision with the sidewall of the apertures when passing through them to the process chamber. The neutralized efficiency can be as high as more than 95%. On the other hand, more than 90% of UV intensity can be eliminated when UV photons pass through the graphite plate to the process chamber. As a result, only energetic oxygen neutral beams bombard the graphene surface for nanoribbon patterning. The beam energy ranging from several to hundreds of electron-volts can be controlled by applying an RF or a DC bias to the bottom electrode. Additionally, as the neutral beams are well aligned, the mask pattern can be transferred precisely to the graphene sheet. To demonstrate the potential of low damage of neutral beam to graphene material, pristine graphene without pattern were irradiated by nitrogen (N2) NB and N2 plasma with the same ICP conditions as follows: chamber pressure of 1.7 mTorr, an RF power
of 250 W, an Ar flow rate of 6.7 sccm, and an irradiation time of 5 min. The generated ion and beam flux under these conditions are approximately 1016 s-1 cm-2 for N2 NB and N2 plasma, respectively. Figure 2 shows the Raman spectra of graphene sheet after irradiations of N2 NB and N2 plasma. Well-defined features of the Raman spectra can be identified: G-band (in-plane vibration of sp2 carbon atoms, ~1585 cm−1); the D-band (defect induced bands, ~1350 cm−1); and the 2D-band (double-resonance band, ~2700 cm−1). We can clearly see that no featured bands of graphene can be observed after irradiation of N2 plasma. On the contrary, the Raman spectrum of the graphene sheet after irradiation of N2 NB shows the featured bands of graphene (G-band and 2D-band) with small D-band. Those results indicate that the NB process is an ultra low damage process for graphene due to low energy ion bombardment and elimination of UV irradiation. 3.2 Structure of Graphene Nanoribbon Array Four nanoribbon arrays with widths of 30, 50, 70, and 100 nm were patterned by EBL and subsequently etched by O2 NBE to obtain GNR arrays (see Experimental Section for details). Figure 3(a) shows the OM images of the GNR arrays after removing hydrogen silsesquioxane (HSQ), a negative-tone resist. We found that the unmasked graphene was etched away cleanly by NBE and only the GNR arrays were left, which clearly have a rectangular shape. Figure 3(b) presents the AFM images of GNR arrays
of various widths after removing HSQ. The AFM images revealed that the GNR arrays can be fabricated uniformly on a large scale using EBL followed by O2 NBE. The profiles of the GNR arrays with various nanoribbon widths measured by AFM are shown in Figure 3(c). The SiO2 under the graphene is also etched when removing the HSQ by using diluted hydrofluoric acid, the height is normalized by the total depth (around 30 nm) as measured by AFM. We can clearly see that the widths of the fabricated GNR arrays are close to the same size as the patterns designed by EBL. The images and figures exhibit uniform patterning of the GNR arrays using EBL followed by O2 NBE for large-scale integrated circuits. 3.3 Raman Spectroscopy of GNR Representative Raman spectra of the GNR arrays are shown in Figure 4(a). Well-defined features of the Raman spectra can be identified as mentioned in section 2.1 and an additional D′-band was observed. Both D- and D′-band are responsible for defect-induced bands and D′-band appears when the GNR width is narrower due to the presence of edges. [16,19,29]. A square graphene sheet with a width of 5 μm shows a symmetric peak with a high intensity (the intensity ratio of the 2D-band to the G-band is higher than ~4) with an almost invisible D-band. These features indicate that high-quality and single-layer graphene was grown by CVD [30]. As shown in Figure 4(a), clear D- and D′-bands were observed because of defect generation at the
GNR edge; no such bands were observed in the bulk sample. No substantial shift in the spectra was observed as the width of the GNR array changed. Figure 4(b) shows the ratio of the integrated intensities of the D- and G-bands, denoted as ID and IG respectively, as a function of the GNR width. The ID/IG was the indicator of the defect level at the GNR edge. Although the ID/IG ratio increases as the GNR width decreases, the ratio is impressively lower than that of the GNRs fabricated by conventional O2 plasma etching [7,16]. Since the graphene is covered by resist, the anisotropic ion bombardment cannot induce such wide edge defect in GNR under the resist. Therefore, the low edge defect should arise from the reduction of UV irradiation damage. Because the UV photons, which are isotropic, can penetrate into the resist and reach graphene nanoribbon especially from the edge where penetration path is relatively short as schematically shown in Figure 5. It was reported that the photon energy of O2 plasma is as high as 9.5 eV [31] which is much higher than graphene C-C bond energy of 4.9 eV [32]. Therefore, in the case of conventional plasma etch, the sp2 C-C bonds at the edge of GNR are destroyed to form wide range of defects (indicated as w in Figure 5(a)) observed as high ID/IG in Raman spectroscopy. On the contrary, the UV photons are almost eliminated in the NB system and therefore only narrow range of defect (as indicated as w′ in Figure 5(b)), contributing to small ID/IG, is induced owing to the C-C bond breakage at the edge resulting from NB
bombardment, which is the nature of etch process. Moreover, the ratio is comparable to that of GNRs prepared by unzipping carbon nanotubes followed by post annealing [13,14], a procedure which is not suitable for large-scale fabrication. It has been reported that post-fabrication treatment is required to repair the high degree of disorder at the edge [14,17,33], and this is what makes the unzipping procedure complicated. The results in Figure 3 strongly suggest that high-quality GNR arrays can be fabricated easily using NBE without post-fabrication treatment. Furthermore, Raman mapping of the etched samples, as shown in Figure 6, further demonstrated the potential of large-scale GNRs fabrication using NBE; no graphene-feature bands were observed in the Raman spectra of the unmasked areas. 3.4 Characteristics of GNR-based FET To evaluate the electrical characteristics of the GNRs fabricated by NBE, a bottom-gate FET with a GNR array of 10 nanoribbons (width: 70 nm) was fabricated (shown schematically in Figure 7(a); an OM image of the FET is shown in Figure 7(b). The transfer curve (drain current Id vs. gate voltage Vg) with an applied voltage between the source and the drain (Vsd) of 0.5 V for the FET is shown in Figure 6c. We repeated the FET fabrication for three times and measured 15 devices. We obtained 10 devices with similar transfer curve as shown in Figure 7(c). The field-effect mobility (μ) was extracted based on the slope ΔId/ΔVg of a line fitted to the linear regime of the
transfer curve using the following formula [21,34,35]: μ=
(ΔI
d
ΔV g )L
nWV sd C g
(1)
where L, n, and Cg are the channel length, number of GNRs, and capacitance per unit area between the GNR and the bottom gate, respectively. We obtained an effective field-effect mobility of more than 200 cm2 V−1 s−1. Although the mobility value is not as high as single crystalline graphene [21,22] that is obtained by mechanical exfoliated method. However, mechanical exfoliation method to obtain graphene is not suitable for very-large-scale integration (VLSI), which is highly anticipated application of graphene, because it is done by hand and the area of graphene is only around several hundreds of micrometer square. This small area of single crystalline graphene sheet is without grain boundaries and different from large are of polycrystalline CVD-grown graphene. Therefore, the mobility of more than 200 cm2 V−1 s−1 is high for CVD-grown GNR due to that CVD-grown graphene is inevitable to contain grain boundaries in channel region, especially in long channel (5μm in this study). Comparing to mechanical exfoliation method, CVD is a VLSI-compatible process, and the area of CVD-grown graphene sheet can reach wafer-size. The result demonstrates high-quality GNRs with low edge defects, which reduces edge scattering for high carrier mobility. Furthermore, we observed high Ion/Ioff current ratio ranging from 103 to 104 at room temperature in our experiment, which is much higher
than other reported results [3]. Such large Ion/Ioff current ratio should contribute from transport gap. M. Y. Han et al. reported that transport gap strongly depends on length of GNR [36]. Since our fabricated GNR is 8μm long (i.e. long channel), which is much longer than that fabricated by other reported approaches. As a result, we considered that such large Ion/Ioff current ratio may result from long channel effect.
4. Conclusions In summary, promising GNR arrays with low edge-defects were demonstrated using EBL and followed by O2 NBE on large-scale CVD-grown graphene. AFM and Raman analysis show that high-quality GNR arrays can be fabricated because NB system can provide low energy and well-controlled neutral beams without UV irradiation damage. A bottom-gated FET having GNR array with nanoribbon width of 70 nm was fabricated. We obtained a high field-effect mobility of ~200 cm2 V−1 s−1 and observed a high Ion/Ioff ratio (103~104) in our fabricated FET. This method is a good candidate for fabricating high-performance GNR-based electronic devices for large-scale integrated circuits.
References
[1] Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos SV, Grigorieva IV, Firsov AA. Electric field effect in atomically thin carbon film. Science 2004;306(5696):666-9. [2] Geim AK. Graphene: status and prospects. Science 2009;324(5934):1530-4. [3] Schwierz F. Graphene transistors. Nat Nanotech 2010;5(7):487-96. [4] Lim WS, Kim YY, Kim H, Jang S, Kwon N, Park BJ, Ahn JH, Chung I, Hong BH, Yeom GY. Atomic layer etching of graphene for full graphene fabrication. Carbon 2012;50(2):429-35. [5] Son YW, Cohen ML, Louie SG. Energy gaps in graphene nanoribbons. Phys Rev Lett. 2006;97(21):216803. [6] Lee H, Kim S, Ihm J, Son YW, Choi HJ. Field-induced recovery of massless Dirac fermions in epitaxial graphene on SiC. Carbon 2011;49(7):2300-5. [7] Wang X, Dai H. Etching and narrowing of graphene from the edges. Nat Chem 2010;2(8):661-5. [8] Han MY, Özyilmaz B, Zhang Y, Kim P. Energy band-gap engineering of graphene nanoribbons. Phys Rev Lett 2007;98(20):206805. [9] Li X, Wang X, Zhang L, Lee S, Dai H. Chemically derived ultrasmooth graphene nanoribbon semiconductors. Science 2008;319(5867):1229-32.
[10] Yang X, Dou X, Rouhanipour A, Zhi L, Räder HJ, Müllen K. Two-dimensional graphene nanoribbons. J Am Chem Soc 2008;130(13):4216-7. [11] Cao X, He Q, Shi W, Li B, Zeng Z, Shi Y, Yan Q, Zhang H. Graphene oxide as a carbon
source
for
controlled
growth
of
carbon
nanowires.
Small
2011;7(9):1199-202. [12] Kosynkin DV, Higginbotham AL, Sinitskii A, Lomeda JR, Dimiev A, Price BK, Tour JM. Longitudinal unzipping of carbon nanotubes to form graphene nanoribbons. Nature 2009;458(7240):872-6. [13] Cataldo F, Compagnini G, Patané G, Ursini O, Angelini G, Ribic PR, Margaritondo G, Cricenti A, Palleschi G, Valentini F. Graphene nanoribbons produced by the oxidative unzipping of single-wall carbon nanotubes. Carbon 2010;48(9):2596-602. [14] Shimzu T, Haruyama J, Marcano DC, Kosinkin DV, Tour JM, Hirose K, Suenaga K. Large intrinsic energy bandgaps in annealed nanotube-derived graphene nanoribbons. Nat Nanotech 2011;6(1):45-50. [15] Bai J, Duan X, Huang Y. Rational fabrication of graphene nanoribbons using nanowire etching mask. Nano Lett 2009;9(5):2083-7. [16] Ryu S, Maultzsch J, Han MY, Kim P, Brus LE. Raman spectroscopy of lithographically patterned graphene nanoribbons. ACS Nano 2011;5(5):4123-30.
[17] Liu L, Zhang Y, Wang W, Gu C, Bai X, Wang E. Nanosphere lithography for the fabriation of ultranarrow graphene nanoribbons and on-chip bandgap tunning of graphene. Adv Mater 2011;23(10):1246-51. [18] Lin MW, Ling C, Zhang Y, Yoon HJ, Cheng MC, Agapito LA, Kioussis N, Widjaja N, Zhou Z. Room-temperature high on/off ratio in suspended graphene nanoribbon field-effect transistors. Nanotechnology 2011;22(26):265201. [19] Bischoff D, Güttinger J, Dröscher S, Ihn T, Ensslin K, Stampfer C. Raman spectroscopy on etched graphene nanoribbons. J Appl Phys 2011;109(7):073710. [20] Yang Y, Murali R. Impact of size effect on graphene nanoribbon transport. IEEE Electron Dev Lett 2010;31(3):237. [21] Shin YS, Son JY, Jo MH, Shin YH, Jang HM. High-mobility graphene nanoribbons prepared using polystyrene dip-pen nanolithography. J Am Chem Soc 2011;133(15):5623-5. [22] Liao L, Bai J, Lin YC, Qu Y, Huang Y, Duan X. High-performance top-gated graphene-nanoribbon
transistors
using
zirconium
oxide
nanowire
as
high-dielectric-constant gate dielectrics. Adv Mater 2010;22(17):1941-5. [23] Lu AY, Wei SY, Wu CY, Hernandez Y, Chen TY, Liu TH, Pao CW, Chen FR, Li LJ, Juang ZY. Decoupling of CVD graphene by controlled oxidation of recrystallized Cu. RSC Adv 2012;2(7):3008-13.
[24] Ohchi T, Kobayashi S, Fukasawa M, Kugimiya K, Kinoshita T, Takizawa T, Hamaguchi S, Kamide Y, Tatsumi T. Reducing damage to Si substrate during gate etching process. Jpn J Appl Phys 2008;47(7):5324-6. [25] Ishikawa Y, Okigawa M, Yamazaki S, Samukawa S. Reduction of palsma-indued damage in SiO2 film during pulse-time-modulated plasma irradiation. J Vac Sci Technol B 2005;23(2):389-94. [26] Samukawa S, Sakamoto K, Ichiki K. High-efficiency neutral-beam generation by combination of inductively coupled plasma and parallel plate DC bias. Jpn J Appl Phys 2001;40(7B):L779. [27] Samukawa S, Sakamoto K, Ichiki K. Generation of high-efficiency neutral beam by using negative ions in an inductively coupled plasma source. J Vac Sci Tech A 2002;20(5):1566-73. [28] Huang CH, Wang XY, Igarashi M, Murayama A, Okada Y, Yamashita I, Samukawa S. Optical absorption characteristics of highly ordered and dense two-dimensional array of silicon nanodiscs. Nanotechnology 2011;22():105301. [29] Eckmann A, Felten A, Mishchenko A, Britnell L, Krupke R, Novoselov KS, Casirashi C. Probing the nature of defects in graphene by Raman spectroscopy. Nano Lett 2012;12(8):3925-30.
[30] Graf D, Molitor F, Ensslin K, Stampfer C, Jungen A, Hierold C, Wirtz L. Spatially resolved Raman spectroscopy of single- and few-layer graphene. Nano Lett 2007;7(2):238-42. [31] Ishikawa Y, Ichihashi Y, Yamasaki S, Samukawa S. UV photon-induced defects and its control in plasma etching process. J Appl Phys 2008;104(6):063308. [32] Erickson K, Erni R, Lee Z, Alem N, Gannett W, Zettl A. Determination of the local chemical structure of graphene oxide and reduced graphene oxide. Adv Mater 2010;22(40):4467-72. [33] Jiao L, Zhang L, Ding L, Liu J, Dai H. Aligned graphene and crossbars from unzipped carbon nanotubes. Nano Res 2010;3(6):387-94. [34] Lee CW, Weng CH, Wei L, Chen Y, Chan-Park MB, C. H. Tsai, K. C. Leou, C. H. P. Poa, J. Wang and, L. J. Li. Toward high-performance solution-processed carbon nanotube network transistors by removing nanotube bundles. J. Phys. Chem. C, 2008, 112(32), 12089-91. [35] Lin MW, Ling C, Agapito LA, Kioussis N, Zhang Y, Cheng MMC, Wang WL, Kaxiras E, Zhou Z. Approaching the intrinsic band gap in suspended high-mobility graphene nanoribbons. Phys Rev B 2011;84(12):125411. [36] Han MY, Brant JC, Kim P. Electron transport in disordered graphene nanoribbons. Phys Rev Lett 2010;104(5):056801.
Figures
Figure 1. Neutral beam system. ICP plasma is generated by applying RF power to 3-turn antenna. Charged species are neutralized and UV photons are eliminated by graphite plate with many apertures. Only neutral beams with controllable energy arrive at the etch surface in etch chamber.
Figure 2 Raman spectra of graphene sheet irradiated by N2 plasma and N2 neutral beam under the same ICP conditions.
Figure 3 a) Optical microscope images, b) atomic force microscope images and c) profiles of the GNR arrays with various nanoribbons.
Figure 4. a) Representative Raman spectra of GNR arrays with various widths. b) Ratio of the integrated intensities of the D- and G-bands (ID/IG) as a function of the
GNR width. Results are shown for GNRs prepared by our method and other approaches.
Figure 5. Illustration of defect generation near the edge of GNR patterned by a) conventional plasma and b) neutral beam.
Figure 6. Raman mapping of the etched graphene sheet. Position A is inside the GNR array (GNR width: 70nm), and positions B and C are outside the array. The Raman mapping demonstrates uniform patterning using EBL followed by O2 NBE.
Figure 7 a) Structure of the field-effect transistor with a GNR array. b) Optical microscope image of the bottom-gate FET device. c) The drain current vs. gate voltage (Id vs. Vg) of the FET with a GNR array of 10 nanoribbons (width: 70 nm). (Vsd = 0.5 V)